Quick summary: Transit teams often track schedule adherence, dwell time, and bunching as three separate metrics to fix one at a time. They’re not separate. They’re symptoms of the same underlying feedback loop, and fixing one in isolation can quietly make another one worse.
What you’ll take away:
- Schedule adherence, dwell time, and bunching are mechanically linked, not independent problems
- A delay at one stop increases dwell time at the next, which increases the delay, which increases dwell time again
- The vehicle behind a late bus tends to catch up, not fall further behind, which is what causes bunching
- Fixing one metric without accounting for the others can shift the problem downstream instead of solving it
- Reliability improves when teams look at the feedback loop, not one dashboard number at a time
Fixed Route is transit service that runs a set path on a published schedule. Ask most transit teams what “reliable service” means, and they’ll point to a dashboard: schedule adherence percentage, average dwell time, a bunching alert. Each number gets its own owner, its own target, and its own fix. That’s a reasonable way to organize a team. It’s not an accurate way to describe how these three things behave on the street.
Schedule adherence, dwell time, and bunching are not three separate problems. They are three visible symptoms of a single feedback loop. Treat them as separate, and a fix aimed at one can make the others worse without anyone noticing until the pattern shows up again a few stops down the line.
How the Feedback Loop Actually Works
The mechanism is well documented in transit operations research, and it’s simpler than the number of dashboards built around it would suggest. A bus that falls behind schedule arrives at its next stop later than riders expected. More riders have accumulated at that stop in the meantime, because more time has passed since the last vehicle came through. More riders boarding means more dwell time. More dwell time means the bus falls further behind. The delay and the dwell time reinforce each other, each one making the next stop’s version of the same problem worse.
This is also exactly what produces bunching. Research on bus service reliability describes the mechanism directly: a bus running a long headway (more time since the previous bus) tends to pick up more waiting passengers than usual, which slows it down further, while the bus behind it, now facing a short headway and fewer waiting passengers, moves faster and catches up. Two buses that started the day evenly spaced end up traveling in a pair, and the gap behind them grows to match. The pattern isn’t a coincidence or a sign of a poorly trained operator. It’s the predictable output of schedule delay and dwell time reinforcing each other along a route.
TCRP Report 30, the Transit Cooperative Research Program’s foundational reference on transit scheduling, treats schedule adherence checks, dwell time data, and running-time data as parts of the same operational picture agencies need to evaluate together, not as separate metrics to be optimized independently. That’s a meaningfully different posture than treating each one as its own KPI with its own dashboard tile.
1 feedback loop, 3 dashboard metrics
Schedule adherence, dwell time, and bunching are commonly tracked separately, but they are mechanically linked.
1 feedback loop, 3 dashboard metrics
Schedule adherence, dwell time, and bunching are commonly tracked separately, but they are mechanically linked.
(Source: Internal framework, based on the cited research)
Why Fixing One in Isolation Can Backfire
Because these three things are mechanically connected, a fix aimed at only one can push the problem somewhere else instead of resolving it.
Padding the schedule to improve adherence can hide the underlying dwell time problem rather than fix it. A route with more built-in recovery time will show a better on-time percentage without the actual boarding process getting any faster, and the extra slack can mask a dwell time issue that keeps getting worse.
Speeding up boarding at one high traffic stops without addressing the schedule gap that created the crowd in the first place can shift the bunching pattern to a different stop further down the route, rather than eliminating it.
Chasing a bunching alert reactively, by holding or short turning a vehicle in the moment, treats the visible symptom at the point it’s noticed rather than the schedule and dwell time dynamics that produced it several stops earlier.
None of these responses are wrong on their own. They’re incomplete when applied without visibility into how the three metrics are interacting on that specific route, at that specific time of day.
Late bus, more riders, more delay
A vehicle running behind schedule encounters more waiting passengers, increasing dwell time and pushing it further behind.
Late bus, more riders, more delay
A vehicle running behind schedule encounters more waiting passengers, increasing dwell time and pushing it further behind.
(Source: Transit service reliability research, ScienceDirect)
A Framework for Looking at Reliability as a Whole
Instead of asking “how do we improve schedule adherence,” “how do we reduce dwell time,” and “how do we stop bunching” as three separate questions, a more accurate starting point is to ask where in the feedback loop a route’s problems are originating.
Where does dwell time variability concentrate? A handful of high-traffic stops driving most of the variability point to a boarding-process fix. Even a small dwell time increase at a high-frequency stop, however, is a different problem than dwell time variability spread evenly across a route.
Is a bunching pattern showing up at a predictable point on the route? If bunching consistently appears a few stops after a known slow segment, the root cause is upstream, not at the point where the bunching becomes visible.
Is schedule padding compensating for a dwell time problem, or for a genuine running-time issue? These require different fixes. Padding that’s covering for boarding delays at specific stops won’t be solved by adjusting total route time.
Are adherence, dwell time, and bunching being reviewed together, by route and by time of day, or only as separate system-wide averages? A system-wide average can look acceptable while masking a specific route or time window where all three are reinforcing each other badly.
Reliability, from a rider’s perspective, is a single experience: did the vehicle show up close to when it was supposed to. From an operations perspective, that single experience is produced by a loop, not a checklist. Teams that monitor these three metrics together, and trace problems back to where the loop starts, tend to make more durable fixes than teams optimizing each number independently.
Short headway, fewer riders, faster trip
The following vehicle faces a shorter headway and fewer waiting passengers, letting it catch up and creating a bunched pair.
Short headway, fewer riders, faster trip
The following vehicle faces a shorter headway and fewer waiting passengers, letting it catch up and creating a bunched pair.
(Source: Transit service reliability research, ScienceDirect)
Frequently Asked Questions (FAQ): Schedule Adherence, Dwell Time, and Bunching
Is schedule adherence, dwell time, and bunching really connected, or just correlated?
They’re mechanically connected. A delay increases the number of waiting riders at the next stop, which increases dwell time, which increases the delay further. That same loop is what causes the following vehicle to catch up and create a bunched pair.
Why does the vehicle behind a late bus tend to catch up to it?
Because a shorter time has passed since the previous vehicle came through, fewer riders have accumulated at each stop, so the following vehicle spends less time dwelling and closes the gap, which is the mechanism that produces bunching.
Can improving schedule adherence alone fix a bunching problem?
Not reliably. Padding a schedule can improve an on-time percentage without addressing the dwell time dynamics that caused the delay in the first place, which can leave the underlying bunching risk in place.
Can improving schedule adherence alone fix a bunching problem?
Not reliably. Padding a schedule can improve an on-time percentage without addressing the dwell time dynamics that caused the delay in the first place, which can leave the underlying bunching risk in place.
Where should a team look first when reliability problems show up?
At where dwell time variability concentrates and where a bunching pattern first becomes visible on the route, since the root cause is often upstream of the point where the problem is noticed.
Should agencies track these three metrics together or separately?
Together, by route and by time of day. A system-wide average for each metric can look acceptable while masking a specific route or time window where all three are reinforcing each other.
Is this feedback loop unique to high-frequency routes?
It’s most visible on high-frequency routes because the timing between vehicles is short enough for small delays to compound quickly, but the same schedule-delay-to-dwell-time relationship applies to any route with enough ridership for boarding volume to vary with wait time.
Want to See Where This Loop Shows Up on Your Own Routes?
Every network has its own version of this feedback loop, often concentrated on a handful of routes or time windows rather than spread evenly across the system. If your team wants a clearer picture of where schedule adherence, dwell time, and bunching are reinforcing each other on your own network, TransLoc’s reporting tools can help connect those metrics rather than viewing them one dashboard at a time.